@martian - yeah, it's problem with the plugin. I'm going to try to fix that come the next Visualize This segment. I hope it's not of too much inconvenience :P
FlowingData Forums » Data Visualization
Visualize This: Poverty Rate By Age in America (Jan 14 to Jan 27)
-
-
@uuilly, very impressive! (I didn't see your submission until after I made my post.) Actually, I prefer the still images to the animation (but others may not think so).
-
@Pat McComb Thanks a lot! You too. I like how yours brings in the notion of total population in each group. Well played...
-
Check out this site
-
Nothing fancy, but I think it portrays the data ... plus a bit of emotional impact (if you can see the background picture).
[attachment=306,70]
-
Attached is my effort, I have taken the supplied data and appended some Crime Statistics from the FBI website, as you can see there is some strong/alarming correlation between Crime and Poverty, hence the headline "Which comes first?"
Interesting the only crime that doesn't show a strong correlation with any age group is "Forcible Rape", I'm sure there are many discussion that could be had as to why that is. The scatter plots are then the overall trend for Total Crime across each states, there are some major differences across the states, as have been shown in previous vizs.Charts created using wonderful Excel, Maps with ArcView, Crime data from FBI.gov
Thanks for the idea Nathan, I've been thinking what to do with the data since it was posted, hence the last minute final viz.
[attachment=308,72]
-
man, i didn't see this competition was going on but it's a great idea Nathan. Congrats to all those who submitted.
-
A slideshare of maps illustrating the difference between the poverty rates of states and the U.S. is at the link below.
http://www.slideshare.net/mstoddard/difference-in-poverty-rates-states-and-us-presentation
I created the maps with many eyes. Go to the link below to see the interactive maps.
http://manyeyes.alphaworks.ibm.com/manyeyes/visualizations/difference-in-poverty-rates-states-a
What jumped out at me is the divide between north and south.
-
Pat -
Both of your charts have an unintended and false indicator. Above the gray band is a conspicuous label saying "US 17% Poverty", which led me for a while to think the gray band itself indicated the 17% benchmark. When I finally saw that nowhere did any values exceed the width of this band, I realized that the gray band just served as the poverty part of the chart, i.e., the negative space, and to the right of the band, in the white, meant not in poverty. I'm still not liking the areal representation and the one dot per 100k residents, especially since the horizontal dimension is not used to show the percentages. Since it's not available graphically, I need to read the numbers.
MDJohnson & Silas Fisher -
I liked the version by Silas better, with one bar to indicate the average and dots to indicate the individual measures. The version by MD was too cluttered with three sets of bars and a single set of dots.
-
Here's my own humble submission. I started with bar charts, and I have a sheet filled with dozens of them. I felt they were too cluttered to show more than one population on the same graph. I did some scatter charts of one age group vs. another, but found nothing more than a little noteworthy. I finally settled on an approach using lines and markers.
I'm not really satisfied with the colors, but I like the lines and markers for being relatively less cluttered than other approaches might be. I used a dashed line for US average rather than another point in the middle of the sort order (or alternatively at either end of the list).
I sorted by the overall rate. Seemed to make the most sense. I toyed with leaving this line off the chart altogether (other than the sort order), but I relented at the last moment. I still think it adds more clutter than insight.
There is no explicit North-South relationship in my chart, other than perhaps noting that Mississippi is at one extreme and New Hampshire the other.
What is most striking to me, and most disturbing, is that the population with the largest percentage living in poverty is our children. Pretty sad, for supposedly the richest nation in the world. I suppose the lower percentage of the elderly living in poverty might illustrate the lower life expectancy of those who spend a life in poverty.
[attachment=313,73]
-
My vote:
Uuilly's overall rate animation. It's the only one that made a really interesting and salient fact just pop out:
Washington DC has an enormously higher overall rate (worse for elderly and children) than its surrounding states. It juts out from the visualization as this isolated tower of suffering in what is otherwise a region of soothing and non-worrisome blue.
To me, that's the hallmark of a great visualization. It helps turn mere data (the raw numbers) into knowledge ("The children and elderly in Washington DC have a serious poverty problem as compared with those in the surrounding states").
Data is, well, boring. Knowledge is motivational. Saying "DC's poverty rate is no worse than the national average" doesn't motivate anyone to do anything. Saying "Kids in DC are more than twice as likely to live in poverty than those in neighboring New Jersey or Maryland" at least begs people to ask "well, why is that?"
So, for giving me a compelling new piece of knowledge, Uuilly gets my vote.
-
I did like UUilly's as well .. however, I wanted to see all of the data represented on a single graph. Otherwise, if it was for a single demographic, I think the tool he chose to use was the best choice. However, as a bit of a purist, I appreciate the use of commonly available tools (ex. MS Office products, since that's what I have to use).
I too like Sila's chart. I only wish he would have used a horizontal indication for the US average for a simpler comparison of the US total versus each independent state.
@Jon Peltier ... In hindsight, I agree that the bar charts are too cluttered ... or at a minimum, the colors should not be so contrasting. I like your chart, but wanted to see the vertical dotted lines labeled for people less familiar with the data set.
Anyway, fun stuff.
-
@cloister The DC data has a hidden variable. If you pull out the core part of any metropolitan area it's going to also swamp the average across a whole state.
With that said: there are serious problems in DC, chief among them its embarrassingly bad educational system (tho new chancellor Rhee is apparently really turning things around).
-
@mrflip
That is a valid observation. However, considering that DC is its own political entity with its own executive management separate from the commonwealths of MD and VA, the observation would seem to carry some weight.
The data itself is grouped by politically independent geographies, so in this instance I think the conclusion is still meaningful.
You're right, though, that generally speaking we have to watch for artifacts introduced by varying resolution in the data, which is what we have here.
-
Hi, this is my first post ever on this forum.
For this graph I just wanted to tell a story and put an accent on one problem, instead of giving back a tool able to compare every State's value. That's why the focus of the graph is rather visualizing distribution and concentration of values than exact data representation.Basically it's a scatter plot, the colored areas show the distribution of the various points (I added a GDP per capita data set for every State).
The USA map can be used to show the geographic distribution of wealth but I really didn't have time to do it, I will add it as soon as I have some.[attachment=322,74]
-
"Non ho più soldi, sono rimasto in mutande" (I've no more money, I'm in pants).
From this expression that has in "to lose one's shirt "his English equivalent, we wanted to use a way to represent statistical data, different from the usual pie or bar charts.
We wanted to leave in pants The Barbie, pop symbol of American consumerism !
Each polaroid photographs the poverty level of states, the three puppets show the different age groups and the degree of nudity of each shows the rate of poverty.
The assumptions on which we based is that communicate data to an unspecialized public, like a simple citizen, through figures of speech as the analogy or metaphor, enabling closer to its code language can facilitate their understanding.Posted by Mauro Napoli and Daniele Fadda.
Please visit: http://www.densitydesign.org/
[attachment=324,75] [attachment=324,76] [attachment=324,77] [attachment=324,78]
-
The maps I've seen made it easy to pick out the worst regions/places (percentage wise), but they left a lot of questions unanswered. How did a state compare to average? How did a region? How do the percents translate into raw numbers (e.g. are there more poor people in CA or NH than AK given their population differences)? I tried to answer these questions in my visualization.
The left column is the percentage data. The state names are colored by the region (determined by US census regions) and USA is the average. It is still clear that one region dominates the below-average category. However, you can now directly see how states compare to each other.
The right column answers the "how many people" question. It is ranked by population count in the category (data provided by the Kaiser website). The USA entry represents the average number of people in a poverty category per state. Coloring was modified in the second column to highlight states that changed their ranking by a lot (alpha ranges from .25 to .75 with lighter for values that did not move far). I prefer the one with a gray-scale right column, but I'm providing both.
I was surprised how uniformly count and percent rankings did not match (notice, most of the states are of a similar gray value).
These images were produced with the Stencil visualization system, followed by Photoshop to add the reference marks.
Notes:
1) The population count column was originally in color. The natural value difference in the colors selected actually caused confusion (more time could yield a better color scale). It does help show that the geographic distribution of the poverty headcount is not as uniform as the distribution of percents, which is interesting but I didn't think completely justified retaining the colors.
2) Including the raw-data as a fine subscript/underline under each point was examined. It worked in interactive versions, but was either illegible or ugly at static screen resolutions. It would probably work OK in print, as the detail text was about 1/4 the size of the labels.
3) Ordering ties were broken by alphabetical order of state name. This means that some states aren't really worse then their immediate upwards neighbor. Indicating this would be a valuable addition to the schema (maybe joining bars or something).[attachment=325,79] [attachment=325,80]
-
Ok, I'm changing my vote. Luca Masud's work is both visually pleasing and very creative. And, it does a better job of making the salient fact--one egregious outlier--jump right out at you by revealing this new piece of knowledge:
"The richest place in the country shares the nation's highest children's poverty rate with the poorest place in the country."
Wow.
-
Favorite: Lucas Masud
Visually striking and it makes a very interesting fact out GDP and poverty rates blindingly clear.Some more specific comments--
@Lucas: The name fan-out in the middle adds some important context. I wonder if map is worth the space it consumes. It really just contextualizes MS and DC... The "poverty-peak" commentary is good and well stated (and well understated), it left me wondering where other such commentary was though. From an understanding standpoint, most of the data is swallowed up in the big blob. This works for highlighting extremes, but against understanding more detailed information. You can see the "children have it bad" fact, but the other trends are obscured. Maybe the fan-out could have been region coded, or the map more fully employed.@Pat McComb: I like the offset circle blocks to represent the population in/out of poverty. I think it establishes both count and percent well. For improvement, I think the US section dominates the individual states. It makes it hard to compare the values to the average. I would reduce the US one and examine the order of presentation. Alphabetical is a good default ordering, but is it the best for this case?
@Hadley: I like the ordering by total poverty, the axis label could point this out quietly to some benefit. Also, the legend is missing the dark-gray dot key and the NH children's rate dot seems to be missing. Of the two options, I like the firs one better. I found I confused the adjacent states less since each state's markers were closer together. The 2nd plot is interesting, but could use some work to make it more clear.
@whatype: Good job showing both rates and populations. I think that the combination is very important to the story of poverty. The black bars are a bit heavy, and dominate the graphic (first blush, I see a lot of the number 3 being important..when that is really just incidental to the split we're using). I think the geography-sensitive layout was good. Did you try color coding the background circles? For example, by best/worst rage of a given variable? It could help highlight interesting areas.
@uncoveringdata: Props for bringing in unique additional data. It bugs me that the chart title question is not answered in the charts presented.
@silas: The similarity to error bars on your chart took me a minute to get over. I thought it was just a bar chart at first! On further consideration, I like the idea, but maybe something could be done to distinguish (esp. the points in the bar) a bit more. More than most, this chart calls out the fact that the children's rate is always higher than the overall rate.
General:
Lots of these graphics do a good job of highlighting known patterns, but might not be good at helping discover related patterns. For example, just color coding by N/S works well because that is how the data distributes. However, if the pattern were E/W or NE/SW it would be obscured in a N/S coding. As a presentation technique, highlighting known patterns is good, but as an exploration technique I think the literal geographic layout or more coarse groupings do a better job. -
Hi Jacottam, you are perfectly right. As for the bottom USA map I already wrote in my entry that I didn't have the time to include the data in it, showing the geogaphic distribution. As it is the map is perfectly useless.
There's another thing I would change: the Washington D.C. peaks are linked with the other but that line represents no data at all! Maybe I should divide them from the rest of the "blob" or putting in the blob the real scatter plot points (and making the blobs much more transparent).
By the way I think that the better way to relate every State to each other clearly is a bar chart (at least with the data set give. Maybe adding GDP or other data I could have worked out something else). That's why I rather chose to tell a story instead of drawing such a tool.
Reply »
You must log in to post.